The Training Process of Many Deep Networks Explores the Same Low-Dimensional Manifold

Abstract

We develop information-geometric techniques to analyze the trajectories of the predictions of deep networks during training. By examining the underlying high-dimensional probabilistic models, we reveal that the training process explores an effectively low-dimensional manifold. Networks with a wide range of architectures, sizes, trained using different optimization methods, regularization techniques, data augmentation techniques, and weight initializations lie on the same manifold in the prediction space. We study the details of this manifold to find that networks with different architectures follow distinguishable trajectories but other factors have a minimal influence; larger networks train along a similar manifold as that of smaller networks, just faster; and networks initialized at very different parts of the prediction space converge to the solution along a similar manifold.

References in corpus (6)

• Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift
• Distilling the Knowledge in a Neural Network
• Understanding deep learning requires rethinking generalization
• Spectrally-normalized margin bounds for neural networks
• An Investigation into Neural Net Optimization via Hessian Eigenvalue Density
• Mean-field theory of two-layers neural networks: dimension-free bounds and kernel limit

Cited by in corpus (1)

• Circuit design in biology and machine learning. I. Random networks and dimensional reduction